The annual stocktake is one of those practices that survives more on tradition than merit. It shuts the plant, consumes everyone for two days, produces a large variance nobody can explain, and delivers accurate records for approximately one week. Then accuracy decays for fifty-one weeks until the next one.

What’s Wrong With the Annual Approach

The timing is the core problem. Errors made in February are discovered in December, by which point the cause is untraceable. You get a number and no information about how it arose, which means nothing gets fixed and the same variance appears next year.

It also encourages a particular kind of blindness. Because everyone knows records will be corrected at year end, discrepancies during the year get tolerated. The stocktake becomes a reset button rather than a check.

And the count itself is often poor. Two days of rushed counting by people doing it once a year, under pressure to finish, produces errors of its own — sometimes enough to make the corrected figure worse than the original.

How Cycle Counting Differs

Instead of counting everything once, you count a small selection continuously. High-value or fast-moving items several times a year; slow, low-value items annually. Nobody stops working, and every item gets checked at a frequency proportional to how much its accuracy matters.

The real advantage isn’t the counting — it’s the diagnosis. A discrepancy found within weeks can be traced. Recent transactions are still in memory, the paperwork exists, and the cause is usually identifiable. Root causes get fixed instead of accumulating.

Setting Frequencies

Base them on value and movement. High-value fast-movers might be counted monthly or quarterly, mid-range items twice a year, low-value slow-movers annually. Add items with a history of discrepancy regardless of their value class — recurring variance is a signal that something in the process is broken for that item.

Aim for a workload of an hour or two a day rather than periodic pushes. Sustainability matters more than intensity; cycle counting programmes fail when they’re too ambitious to maintain.

Making the Transition

Start with a clean baseline — one final full count — then begin cycling. Run both for a year if your auditors require it, which many will until the cycle counting programme has demonstrated coverage and accuracy.

Most auditors accept cycle counting in place of a full stocktake given adequate documentation of frequency, coverage, and variance investigation. Confirm the specific requirements with yours early rather than assuming, since the answer shapes how you design the programme.

Measure the Right Thing

Count accuracy — the percentage of counts matching the record within tolerance — is the headline. But track variance causes as well, categorised: transaction errors, unrecorded scrap, misplaced stock, receiving errors, theft. The distribution tells you what to fix. Accuracy improves as a consequence of fixing causes, not as a result of counting more.

ticktick.ai schedules counts by value and movement class, flags items with recurring variance, and tracks discrepancy causes over time.